This Nvidia partner claims it can reduce data center energy consumption by 50% as the AI surge puts a strain on the power grid.

• Billions of dollars are being invested in data center infrastructure by Big Tech companies such as Microsoft, Alphabet, and Meta to support generative AI. However, this has resulted in a surge in energy demand due to the proliferation of data centers.
• Sustainable Metal Cloud claims that its immersion cooling technology is 28% less expensive to install compared to other liquid-based alternatives and can decrease energy consumption by 50%.
• Sustainable Metal Cloud's platform allows for high-density hosting of GPUs, which is necessary for platforms like Nvidia's Grace Blackwell, as stated by Tim Rosenfield, co-founder and co-CEO of Sustainable Metal Cloud.

The surge in demand for powerful processors and energy to cool data centers is fueling the growth of the artificial intelligence industry.

Sustainable Metal Cloud, a data center company, has the opportunity to expand its "sustainable AI factories" made of HyperCubes in Singapore and Australia.

The HyperCubes are equipped with Nvidia processors and are immersed in polyalphaolefin, a synthetic oil that efficiently dissipates heat. The company claims that its platform can reduce energy consumption by up to 50%, compared to the traditional air cooling technology commonly used in data centers.

Sustainable Metal Cloud's new hosting solution allows for high density hosting of GPUs, which is necessary for platforms like Nvidia's Grace Blackwell, as stated by Tim Rosenfield, co-founder and co-CEO of the company, regarding the recently announced AI graphics processors in March.

The Singapore-based company stated that its immersion cooling technology is 28% less expensive to install compared to other liquid-based alternatives. The HyperCubes are designed to fit into any data center and can be easily deployed in unused spaces within existing data centers.

"Being containerized allows us to quickly deploy our solution anywhere and respond to customer demand by opening up new availability zones," said Rosenfield.

He said SMC is expanding into other markets like Thailand and India.

As a result of the transformative impact of AI, both governments and businesses have experienced a surge in demand for data centers.

In countries like Singapore, where SMC is headquartered, there is a push to reduce energy consumption in data centers in order to support the country's AI ambitions, with a commitment of more than 500 million Singapore dollars ($379.7 million).

ST Telemedia Global Data Centres, a Temasek-backed data center operator in Asia, has provided funding to Sustainable Metal Cloud.

SMC is currently seeking $950 million in funding for its data center expansion beyond Singapore, according to Bloomberg, based on information from sources familiar with the matter.

Liquid cooling picking up pace

As performance demands rise, overheating risks become more significant, prompting technology companies to explore liquid cooling for their data centers.

Liquid cooling can theoretically allow data centers to pack more servers into their existing footprint by eliminating the need for large aisles to circulate cooled air, which can prevent overheating and equipment failure.

In June, Giordano Albertazzi, CEO of Vertiv, a digital infrastructure provider, stated to CNBC that the adoption of liquid-cooling could increase significantly in 2024. Vertiv offers thermal management solutions that include hybrid air- and liquid-cooling, as well as fully liquid-cooled data centers.

At Computex Taipei in June, Jensen Huang, CEO of Nvidia, endorsed Supermicro's direct liquid-cooling technology, stating that it reduces energy consumption and lowers costs in AI data centers.

In June, Charles Liang, CEO of Supermicro, informed CNBC that liquid cooling offers higher power efficiency, resulting in improved performance, reduced pollution, and lower energy expenses.

According to Rosenfield of SMC, challenges persist in implementing liquid cooling, despite the excitement surrounding it.

According to Rosenfield, most data centers are not equipped to handle any type of liquid, including immersion and direct chip cooling. The market is still trying to determine the best approach, and there will likely be multiple solutions.

Albertazzi from Vertiv stated that despite the transition to high-density AI data centers, there will still be a significant amount of air cooling that will continue to occur.